Having the best people is vital to any company. However, hiring the best staff is difficult — you can’t know if your hiring decisions are right until several months (sometimes years) after the hire has been made.

Is your recruitment process letting you down by causing you to make the wrong hiring decisions? How can you design your recruitment to maximise the likelihood that you make the right decisions?

The answer, as often, lies hidden in the data.

Deciding which applicant has the highest potential for future career success is not an easy task.

Today, many hiring decisions are based on haphazard gut feelings. More structured approaches are often nothing more than ticking a few boxes. In contrast, Marketing, Finance, and Insurance practices have long relied on data analytics to help inform their decisions.

So what would such a data-driven approach look like in recruitment?

To answer this question, let’s explore the recruitment practice of a UK-based company hiring MBA graduates. Their recruitment consists of 2 rounds of interviews.

Round 1 is conducted on campus and consists of a 30-minute behavioural interview followed by a 30-minute case interview. Successful candidates are then invited to the UK for a Round 2 Assessment Day which consists of a behavioural interview, a case interview, a group exercise, and a presentation. All six interviews across the two rounds carry the same weight.

At the end of the Round 2 Assessment Day, interviewers come together to recap their interviews and jointly decide which candidates get offers and which get turned down.

Figure 1: Recruitment process of an example company

Let us start by understanding how recruitment decisions are made. The question we want to answer is “How likely will a candidate get an offer, given the outcomes of her interviews?” We do this by calculating the likelihood of getting an offer, given strong interview results, and the likelihood of not getting an offer, given weak interview results. For simplicity, in Figure 2 we show the results of this analysis by ranking the predictive strengths.

Figure 2: Relative predictive strength of each interview

Here, we see that the group interview is the strongest predictor of getting an offer. In other words, candidates who did well in their group interview tend to get an offer, while those who did not do well tend to be turned down.

This insight is powerful because we can know the likely outcome of the Assessment Day BEFORE the interviewers come together to jointly make the decision. Additionally, since the recruitment process did not intentionally overweigh the group interview, the interviewers must have been instinctively placing greater emphasis, and hence basing their decisions, on it.

But are their instincts correct?

If we want to optimise the likelihood of hiring decisions leading to great employees, then our prediction analysis should focus on candidates who become successful employees, rather than those who simply get an offer.

In this example, “successful hires” meant those employees who are performing at or above expectations 24 months after their initial start date. With some performance data, we can re-run this analysis to predict performance. Figure 3 shows the results.

Figure 3: Predictive strength of each interview on candidates likely to perform

Two interesting observations:

When discussing the results of this analysis, one interviewer told us, “We don’t purposely over-weigh the group exercise, but I do admit it’s my favourite interview. It’s the only time I get to see candidates side-by-side, working together.”

This may be the interviewer’s favourite interview, but it is not the most effective in predicting success.

Notably, depending on how the group interview is designed, it can favour extraversion or introversion, so it’s worth being aware where your individual biases lie.

What about the two behavioural and case interviews? One difference between Round 1 and Round 2 interviews is the interviewers themselves. Round 1 interviews were conducted by a dedicated recruitment team who interview all candidates year on year. Round 2 interviews by contrast were conducted by interviewers who were available on the day. “Last minute cancellations are not uncommon leaving recruitment coordinators scrambling to find replacements”, said the department head. This lack of consistency affects the quality of the interviews, and hence, affects the accuracy of hiring decisions.

Armed with these facts, this company set out to redesign its recruitment model. Here are three of the changes they made:

This example illustrates how data analytics can surface flaws that have been hindering a company’s performance for years. It uncovers hidden biases and provides confidence for change.

The analysis described here was relatively simple, requiring only techniques familiar to every math and statistics student. But the value to the business is tremendous — the Head of the department said “A handful of high performers make or break my business — having more is our competitive edge.”

Using data to inform business decisions has long been prevalent in many functions. These functions regard data as sacred, going to great lengths to collect and analyse it.

The recruitment function should be no different.

Irrational biases can lurk anywhere in your people processes.

1. Pick an area where you’re not satisfied with the results or identify commonly held beliefs that you feel are detrimental. This could be recruitment, promotions, retention, referrals, performance, employee satisfaction surveys - no HR or Recruitment topic is safe.

2. Precisely map the decision-making process that leads to those results.

3. Work with a statistician to translate facts into data and run regressions to uncover which process step is faulty. If you don’t have enough people-power in-house to perform this kind of analysis, we at Pirical can help.

If you would like to get started with this analysis in your firm, get in touch — it’s what we do best.

Subscribe